20 Years of Forecasting Air Quality in Georgia. Michael E. Chang

Transcription

1 20 Years of Forecasting Air Quality in Georgia Michael E. Chang

2 Atmospheric Science Ecology Sustainable Engineering Earth & Environmental Science Ocean Science Sustainability Transitions

3

4 Atlanta s Air Quality 1996 vs. Climate 1-hour Average O3 (ppm) Observed Predicted Apr 15-Apr 29-Apr 13-May 27-May 10-Jun 24-Jun 8-Jul 22-Jul 5-Aug 19-Aug 2-Sep 16-Sep 30-Sep 14-Oct 28-Oct Olympics, 19 Jul 4 Aug Unhealthful

5 1. Health alert / management 2. Episodic controls 3. Enhance field studies 4. Improve public awareness and education 5. Manage public / media relations 6. Voluntary controls Why Forecast?

6 Why Forecast? 1. Better Health Management Pollution Exposure Avoid Limit Treat Sensitive Populations Young and Old Patients with preexisting diseases Outdoor workers Athletes

7 Why Forecast? 2. Episodic vs. Comprehensive Controls NAAQS Episodic Control Ozone (ppmv) Comprehensive Control 5/1 5/8 5/15 5/22 5/29 6/5 6/12 6/19 6/26 7/3 7/10 7/17 7/24 7/31 8/7 8/14 8/21 8/28 9/4 9/11 9/18 9/25 Date

8 Why Forecast? 3. More Efficient Field Studies Identify when and where interesting events occur Conserve resources (aircraft and other expensive sampling technologies) Real-time adaptive modifications to plans and protocols Augusta Chronicle

9 Why Forecast? 4. Public Awareness and Education 5. Public / Media relations 6. Voluntary Controls Provide information Reinforce message Effective use of broadcast and print media

10 After the Games: Developing New Forecasting Tools & Methods Past history Climate Database query Synoptic & mesoscale meteorological analysis Subjective (surface obs., maps, soundings, satellite) Objective (Classification and Regression Tree, Nearest Neighbor) Models Regression Neural Net Prognostic Group & Individual learning (Internet based) Accessibility Training Consensus building Forecast archive Performance measures

11 First to adapt grid models for forecasting Journal of the Air & Waste Management Association Nov2000, Vol. 50 Issue 11, p JOURNAL OF THE AIR & WASTE MANAGEMENT ASSOCIATION Volume: 51, Issue: 8; Pages: ; AUG 2001 Application of the Urban Airshed Model to Forecasting Next-Day Peak Ozone Concentrations in Atlanta, Ozone predictions Georgia in Atlanta, ATMOSPHERIC Georgia: POLLUTION Analysis RESEARCH of Volume: 1, Issue: 4; Pages: ; Oct the 1999 ozone season Chang, Using Michael synoptic and classification Cardelino, Carlos to evaluate an operational Cardelino, C; Chang, M; St John, J; Murphey, B; Cordle, J; Ballagas, AIR POLLUTION air Abstract: R; Patterson, MODELING quality forecasting Twenty-four L; AND Powell, ITS to forty-eight-hour APPLICATION system K; Stogner, XVIII in Atlanta ozone J; air quality forecasts are Developments in Environmental increasingly Science; being used Volume: in metropolitan 6; Pages: ; areas 2007 Zimmer-Dauphinee, S to inform the public about potentially Hu, Yongtao; harmful Chang, air quality Michael conditions. E.; Russell, The forecasts Armistead are also G.; behind and Forecasting "ozone action day" programs in which public and private sectors are Abstract: Ozone Odman, prediction and M. has PM2.5 Talat become in southeastern an important activity US in encouraged or mandated to alter activities that contribute to the many U.S. ozone nonattainment areas. In this study, we describe formation of ground-level ozone. Presented here is a low-cost Odman, the ozone M. prediction Talat; Abstract: application Hu, program Yongtao; Since in 2006, the of the Urban Chang, Atlanta a team Airshed Michael metropolitan of forecasters Model E.; (UAM), and area in and Georgia (USA) has been an Eulerian 3- Russell, analyze Armistead the performance using dimensional G. the of high-resolution this program air during quality the forecasting 1999 ozoneforecasting season. aid for From making May ozone to September, (O-3) and a team fine particulate of 10 air matter (PM2.5) system (Hi-Res) as an photochemical-transport grid model for generating next-day peak ozone concentration forecasts. During the summer of 1997, nextday peak ozone concentrations in Atlanta, GA, were predicted both by a quality regulators, forecasts. meteorologists, Here, we and examine atmospheric Hi-Res's scientists O-3 and PM2.5 forecasting Abstract: made a An daily air quality prediction performance forecasting of the for next-day system the Atlanta maximum was metropolitan developed 8-hr average to area aid during the summers of the team of eight forecasters and by the Urban Airshed Model in Forecast ozone operational concentration. ozone and The PM2.5 daily A classificatory forecasting forecast was in evaluation made Atlanta, aided Georgia. approach by two was adopted. The The Mode (UAM-FM). Results are presented that compare the accuracy of linear system regression is based spatial models, on three synoptic a dimensional 3-dimensional classification models air (SSC) quality for weather calendar model, and for Atlanta was used to and the team and the UAM-FM. The results for the summer of 1997 indicate the air no-skill quality ozone prediction cluster persistence and the provides forecasting model. high The days resolution team's into typical performance locally. summer A is weather types of dry preliminary evaluation that the shows UAM-FM that the may system be a better has the predictor potential of of peak ozone

12 First to create smog activity guidelines

13 The use of visual representation technologies by virtual teams to reach consensus in critical tasks: A Field Study of Smog Forecasting Teams Nicholas H. Lurie Chih-Hung Peng Sandra A. Slaughter College of Management Georgia Institute of Technology

14 Air Quality and Forecasts: Measuring Behavioral Impacts Douglas S. Noonan School of Public Policy Georgia Institute of Technology

15 Tropical Storm Allison (2001), Hurricane Gabrielle (2001), and Hurricane Dennis (2005) Three exceptional incidences of increased ground-level ozone and particulate matter concentrations in Atlanta, GA Michael E. Chang Carlos A. Cardelino Billy B. Murphey American Meteorological Society 86 th Annual Meeting Atlanta, Georgia February 1, 2006

16

17 Forecast Performance 1999 & Predicted Observed hr Average Ozone (ppbv) /1 5/15 5/29 6/12 6/26 7/10 7/24 8/7 8/21 9/4 9/18 Date

18 Forecast Performance 1999 & 2013 Predicted Peak 8-hr Average Ozone (ppbv) False Positive 1999 False Negative R² = R² = Observed Peak 8-hr Average Ozone (ppbv)

19 As air quality improves, predicting events becomes harder Worst Air Quality in /4/1999 Worst Air Quality in /30/2013

20 As air quality improves, budgets for air quality forecasting have declined Annual Funding ($) $300,000 $250,000 $200,000 $150,000 $100,000 $50,000 $ Year Ozone Design Value (ppbv)

21 A few takeaways Domestic air quality forecasting is now largely a commodity (there may still be growth opportunities overseas), with decreasing consumer demand. Performance is good enough but getting harder to maintain. There may be a market for lower cost methods, or higher resolution. Research and development ideas and products that leverage forecasting networks still mostly untapped.